VOLGA: perioperative durvalumab plus 3 doses of EV wins both EFS and OS in cisplatin-ineligible MIBC, but pCR of 45% vs 57% in EV-303 leaves the partner question open

Data as of 2026-10-25 (Taipei time). [Fact] statements are sourced; [Inference] statements are this article's judgment.

1. One-line conclusion

[Fact] At VOLGA's first interim analysis, durvalumab plus 3 preoperative cycles of enfortumab vedotin (EV) and 9 postoperative cycles of durvalumab, versus upfront cystectomy, gave EFS HR 0.51 (95% CI 0.38–0.67; P<.0001) and OS HR 0.58 (95% CI 0.40–0.83; P=.0027) (Targeted Oncology). [Inference] MIBC patients who are cisplatin-ineligible or refuse cisplatin now have a second perioperative IO plus EV regimen with OS evidence; but its pCR and EFS figures are both below EV-303, leaving AZ with one selling point, "only 3 doses of EV and less toxicity", which longer follow-up and quality-of-life data must prove.

2. What happened

[Fact] VOLGA is the Phase 3 trial (NCT04960709) reported as ESMO 2026 LBA2, in MIBC patients who are cisplatin-ineligible or refuse cisplatin (ClinicalTrials.gov). 695 patients were randomized 1:1:1 to three arms: arm A received preoperative durvalumab 1500 mg plus EV 1.25 mg/kg every 3 weeks for 3 cycles, then durvalumab for 9 cycles after cystectomy (n=233); arm B added tremelimumab to arm A (n=233); the control arm had upfront radical cystectomy followed by observation or locally approved treatment (n=229). The dual primary endpoints were EFS (blinded independent central review) for each experimental arm versus control; OS was the key secondary endpoint; data cutoff was 2026-03-02 (Targeted Oncology). Tremelimumab was given only 2 times before surgery and 1 time after (Fierce).

[Fact] In the doublet arm, EFS HR was 0.51 (0.38–0.67; P<.0001), median EFS not reached vs 12.1 months, 24-month EFS 64% vs 45%; OS HR 0.58 (0.40–0.83; P=.0027), 24-month OS 78% vs 69%; pCR 45% vs 10% by local review. In an exploratory ctDNA analysis of 112 patients, 60% of those ctDNA-positive at baseline cleared it before surgery; 24-month EFS was 81% for those who cleared and 55% for those who stayed positive. The triplet arm had EFS HR 0.61 (0.47–0.80; P=.0004) and OS HR 0.77 (0.54–1.07; P=.123), not significant at the interim (Targeted Oncology).

[Fact] All-cause grade 3–4 AEs were 64% in the doublet arm, 44% in control and 70% in the triplet arm; 21% in the doublet arm discontinued for AEs; immune-mediated AEs were 33% vs 3% (Targeted Oncology). Treatment-related grade 3–4 AEs were 30% in the doublet and 42% in the triplet (Fierce).

For sponsors: VOLGA shows that a de-intensified design of "short preoperative ADC plus postoperative IO alone" can win OS, so the next MIBC trial need not assume EV-303's full-course EV.

3. Why it succeeded, and why its numbers trail EV-303

[Fact] Across trials, EV-303's Keytruda plus Padcev arm had pCR 57% (control 9%, central review) versus 45% for VOLGA (control 10%, investigator review); 24-month EFS 75% vs 64%, with controls at 39% vs 45%; 24-month OS 80% vs 78%, with controls at 63% vs 69%; EFS risk reduction 60% vs 49% and OS risk reduction 50% vs 42% (Fierce). Fierce argues that because pCR is unaffected by postoperative treatment and both regimens have the same preoperative intensity, it is "a more direct clue" to whether Imfinzi or Keytruda is the better Padcev partner (Fierce). [Inference] The argument holds, but the direction of bias between local and central review is uncertain, so the 12-point pCR gap could be over- or understated.

[Inference] Control-arm performance is the second explanation. VOLGA's control arm had 24-month EFS of 45% and OS of 69%, both above EV-303's 39% and 63% (Fierce), which naturally compresses the relative effect; possible causes are enrolling fitter patients who "refused cisplatin" and allowing the control arm "locally approved treatment" (Targeted Oncology), but the actual postoperative treatment rates by arm could not be found.

[Inference] Treatment intensity is the third explanation. VOLGA gives only 3 doses of EV and 12 cycles of durvalumab in total, whereas EV-303 gives 9 cycles of Padcev before and after surgery and up to 17 cycles of Keytruda (Fierce); de-intensification bought lower treatment-related toxicity, grade 3–4 at 30% for VOLGA versus grade ≥3 at 45.5% for EV-303 (including 2 deaths) (Fierce). The definitions differ and should not be subtracted directly, and the price may be EFS. The triplet arm's OS was not significant, and the presenter also suggested the extra toxicity may have offset the benefit (Targeted Oncology); [Inference] for older, cisplatin-ineligible patients, adding CTLA-4 is not worth it.

4. Regulatory and commercial implications

[Fact] Keytruda plus Padcev won the cisplatin-ineligible MIBC indication on the basis of EV-303 in 2025-11, and Imfinzi plus chemotherapy won perioperative approval in cisplatin-eligible MIBC in 2025 (Fierce). [Inference] EV-303 won approval with significant EFS and OS (Fierce), but the two trials differ in control arm (VOLGA allowed locally approved postoperative treatment, Targeted Oncology) and in pCR review (Fierce), so the precedent applies only by analogy. On that basis, VOLGA also met both EFS and OS, so its chance of approval after filing is reasonable; but it will not change the first choice. In practice clinicians will choose between two IO plus EV regimens, and the deciding factor will be cumulative exposure to EV neuropathy and skin toxicity. If AZ can show quality-of-life and postoperative completion data, it will persuade for older patients. AZ's R&D chief also made quality of life the main argument (Fierce). For the NMPA, this article found no AZ filing plan in China.

5. Recommendations for the next trial and OncoLattice (GU oncology)

[Inference] First, use ctDNA to guide de-escalation or escalation. VOLGA's exploratory data show 24-month EFS of 81% in patients who cleared ctDNA before surgery and 55% in those who stayed positive (Targeted Oncology); next-generation MIBC trials could use preoperative ctDNA to decide whether to continue IO after surgery or to select bladder-preservation candidates. Sponsors should mandate sampling at baseline, before surgery and after surgery, and settle testing logistics in advance.

[Inference] Second, design a bladder-preservation arm. With IO plus EV pCR already at 45–57% (Fierce), "no cystectomy after treatment" is the next big question; a non-inferiority bladder-preservation trial screened by cT0 (TURBT plus imaging plus ctDNA) is recommended, at sites able to perform repeat cystoscopy and MRI VI-RADS reading.

[Inference] Third, site strategy and control arms. MIBC patients usually see a urologist first, so OncoLattice should prioritize hospitals whose urology departments have multidisciplinary teams and can move patients from neoadjuvant therapy to surgery within 6–8 weeks. VOLGA's control arm allowed locally approved treatment (Targeted Oncology), which strengthened it; new trials using IO plus EV as the control should specify postoperative treatment in advance. Finally, pCR should always be read by central pathology review, to avoid another case where VOLGA and EV-303 cannot be compared directly (Fierce).

6. Open questions and what to watch

Central-review pCR, the actual cystectomy completion rate by arm and postoperative treatment use will most likely appear in the LBA2 full paper or later congress presentations; for now there are only abstract reports from ESMO 2026 (10/23–27, Madrid) (Targeted Oncology), and the journal and date have not been announced. Mature OS, final triplet OS and older-patient subgroup results will come from later analyses of NCT04960709, which ClinicalTrials.gov lists as "active, not recruiting" (ClinicalTrials.gov). AZ's FDA filing and decision dates have not been announced; in the Fierce report, AZ's R&D chief said more mature long-term survival data are needed to assess the triplet (Fierce).

The next thing to watch is VOLGA's central-review pCR: if it catches up with EV-303's 57%, the PD-L1 versus PD-1 partner debate falls away, and sponsors can choose an IO backbone on toxicity and dosing convenience alone.

Evidence limitations

The VOLGA versus EV-303 comparison is cross-trial, with different pCR review (VOLGA investigator review, EV-303 central review, Fierce), and the 695 randomized patients differ from the 712 listed on ClinicalTrials.gov (the latter includes the safety run-in, ClinicalTrials.gov).

Advice for readers: sponsors using VOLGA figures to set control-arm assumptions for MIBC trials should wait for the full paper's central-review pCR and postoperative treatment rates by arm before recalculating sample size; until then, use EV-303's central-review figures as the conservative baseline (Fierce).

Note: differences from the original relay

The relayed figures match the sources. Additions: VOLGA's pCR is by local review and EV-303's by central review; the 64% grade 3–4 AE figure is all-cause and the 30% figure is treatment-related, and the two should not be mixed.